阅读说明：本技术 烟草仓库智能化电气火灾预警系统及方法 (Intelligent electric fire early warning system and method for tobacco warehouse ) 是由 喻民军 黄国忠 欧盛南 陈小龙 于 2021-05-25 设计创作，主要内容包括：本发明公开了一种烟草仓库智能化电气火灾预警系统及方法,全面、实时关键参数监测与分析,由于采用了多因耦合综合预警,消除了以往单一因素阈值不精确等问题；本发明能够根据实时监控的火灾风险关键影响因素数据,通过所建立的基于模糊综合评判理论和BP神经网络技术电气火灾综合风险预测模型进行综合分析,确定风险发生的可能性和风险后果严重程度量化值,最终确定电气火灾综合风险水平数值；并且本发明系统具有自学习和自我完善功能,可达到高精度和高可靠性的风险预测目的。(The invention discloses an intelligent electric fire early warning system and method for a tobacco warehouse, which can comprehensively monitor and analyze key parameters in real time, and eliminate the problems of inaccurate threshold value and the like of the conventional single factor due to the adoption of multi-factor coupling comprehensive early warning; according to the invention, the comprehensive analysis can be carried out according to the fire risk key influence factor data monitored in real time through the established electric fire comprehensive risk prediction model based on the fuzzy comprehensive judgment theory and the BP neural network technology, the risk occurrence possibility and the risk consequence severity quantitative value are determined, and the electric fire comprehensive risk level value is finally determined; the system has self-learning and self-improvement functions, and can achieve the purpose of risk prediction with high precision and high reliability.)

1. The utility model provides an intelligent electric fire early warning system in tobacco warehouse which characterized in that includes: the system comprises an external environment parameter detection module, an electrical equipment parameter module, an electrical fire early warning server, a central control system and a risk information display and push module;

the electrical fire early warning server is respectively connected with the external environment parameter detection module, the electrical equipment parameter module and the central control system; the risk information display and push module is connected with the central control system;

the external environment parameter detection module is used for acquiring environment temperature data of the electrical system; the electrical equipment parameter module is used for acquiring voltage, current and power data of an electrical system; the electric fire early warning server is used for processing early warning information and calculating risk level of the acquired data; the central control system is used for controlling the operation of the whole early warning system; and the risk information display and push module is used for presenting, pushing and releasing the fire early warning information.

2. The tobacco warehouse intelligent electric fire early warning system according to claim 1, wherein the external environment parameter detection module and the electric device parameter module both perform data transmission through a WIFI or RS485 network transmission mode.

3. The tobacco warehouse intelligent electrical fire early warning system as claimed in claim 1, wherein the risk information display and push module presents, pushes and issues fire early warning information in an APP, short message or PC-side audible and visual alarm manner.

4. The intelligent electric fire early warning system for tobacco warehouses according to claim 1, wherein the external environment parameter detection module comprises a temperature sensor and a linear temperature sensor; the electrical equipment parameter module comprises a current-voltage transformer.

5. The tobacco warehouse intelligent electric fire early warning system as claimed in claim 4, wherein the distribution positions of the temperature sensors, the line temperature sensors and the current-voltage transformers are divided into areas according to the functional units of the tobacco distribution center.

6. The early warning method of the intelligent electric fire early warning system for the tobacco warehouse according to any one of claims 1 to 5, which comprises the following steps:

s1, collecting current, voltage, environment temperature and line temperature data of the electric fire early warning system;

s2, carrying out fuzzy comprehensive evaluation on the collected data to obtain a learning sample;

s3, constructing a BP neural network model, and dividing the learning sample into a training set and a testing set according to a proportion;

s4, inputting the training set into the BP neural network model for training, and performing effect test by adopting the test set to obtain a trained fire early warning model;

and S5, inputting the data collected in the S1 into a fire early warning model to predict the electrical risk level in real time, and generating an electrical risk level report.

7. The intelligent electric fire early warning method for the tobacco warehouse according to claim 6, wherein the S2 is specifically:

s2.1, determining an evaluation interval corresponding to a risk grade comment set and a risk grade of early warning according to a fuzzy comprehensive evaluation method;

s2.2, selecting the data acquired in the S1, generating a 5-cluster data set according to the condition that the mass center number is equal to 5 through a k-means clustering algorithm, and selecting a point set closest to the cluster center through a k nearest neighbor algorithm;

and S2.3, comprehensively considering factors of current, voltage, temperature and linear temperature based on multi-factor coupling, and acquiring high-risk and extremely high-risk data situations, namely learning samples, from the point set through a cable temperature rise loading coupling experiment.

8. The tobacco warehouse intelligent electric fire early warning method as claimed in claim 7, wherein the risk level comment set includes an extremely low risk, a medium risk, a high risk, and an extremely high risk; the evaluation interval corresponding to the risk level comment set is set as follows in sequence: 0-20, 20-40, 40-60, 60-80 and 80-100.

9. The intelligent electric fire early warning method for the tobacco warehouse according to claim 6, wherein the S4 is specifically:

s3.1, carrying out normalization processing on the training set data, and then inputting the processed training set into the BP neural network model for training;

s3.2, selecting a Sigmoid function as a BP neural network model activation function, and inputting the processed training set into the BP neural network model to perform initialization weight, forward transmission of hidden layer calculation and output layer calculation and reverse transmission of weight update;

and S3.3, performing effect test by adopting the test set, and finishing training when the output result of the BP neural network model reaches the acceptable level of the test set to obtain the trained fire early warning model.

Technical Field

The invention belongs to the technical field of fire early warning, and particularly relates to an intelligent electric fire early warning system and method for a tobacco warehouse.

Background

For the tobacco industry, safe operation is a non-negligible problem, and especially the safety of electrical systems in industrial production plants, storage places and office places is of great importance. In recent years, casualties and property loss due to frequent occurrence of fire accidents caused by electric lines and equipment at home and abroad are difficult to count. In view of the burstiness and the concealment of the electrical fire, the electrical fire monitoring system is widely applied to the safety production supervision process of the tobacco industry.

The existing electric fire monitoring system in the tobacco industry mainly has the following problems to be solved urgently:

1) the fire monitoring systems are accident emergency type alarm systems instead of early warning systems. The issuing of the alarm information of the electric fire monitoring system adopted by the tobacco enterprise at present is established on the basis of detecting the occurrence of smoke or flame, and not an early warning system for serious fire hidden danger when fire does not occur, but only can play the role of accident emergency and fail to achieve the purpose of prevention in advance.

2) The monitoring system with the primary early warning function is single parameter monitoring, the warning limit value is not dynamically changed, and the false warning rate is high. At present, electrical systems in tobacco warehouses, plants and related office buildings face the phenomenon of gradual aging along with the lapse of time, enterprises take measures of regularly monitoring related data such as current or temperature and the like, and once the related data exceed a limit value, the regulation and the like are carried out to carry out prevention and control on fire conditions in advance, so that a certain effect is achieved, but the technology fails to consider the complexity and the contingency of the generation of the fire conditions, and the data monitoring is a data isolated island, so that the accuracy is poor, and the false alarm rate is high.

3) The monitoring system operation state is influenced by human factors to a high degree, and timely emergency response cannot be realized. Most of the current fire monitoring systems are set to be in a manual state by using units or personnel due to high false alarm rate and the like, and are managed by security personnel on duty, the security personnel on duty are usually dispatched by security companies affiliated outside enterprises, the security personnel on duty have high liquidity due to the reason that labor wages are increased in recent years and the like, the training on duty of most of the security personnel on duty is insufficient, the daily management and emergency response are poor, when a fire really happens, the optimal on-site emergency processing time for eliminating the fire is only 3-5 minutes, and under the condition of manual alarm, the most accurate and rapid response within 3-5 minutes by the quality of the current security personnel can not be basically realized.

In summary, the existing electrical fire monitoring system generally adopted in the tobacco industry has many problems, such as low alarm accuracy, and occurrence of situations of false alarm, missing alarm and even no alarm frequently; the comprehensive fire hazard risk level prediction cannot be carried out according to the real-time running state detection data of the system, the emergency treatment effect of accidents can only be achieved, the purposes of early warning and early prevention cannot be achieved, the influence of human factors of operators and users is large, and the timely emergency treatment and treatment effects of the accidents cannot be realized.

Disclosure of Invention

In order to solve the technical problems, the invention provides an intelligent electric fire early warning system and method for a tobacco warehouse, which comprises the following steps: the system comprises an external environment parameter detection module, an electrical equipment parameter module, an electrical fire early warning server, a central control system and a risk information display and push module;

the electrical fire early warning server is respectively connected with the external environment parameter detection module, the electrical equipment parameter module and the central control system; the risk information display and push module is connected with the central control system;

the external environment parameter detection module is used for acquiring environment temperature data of the electrical system; the electrical equipment parameter module is used for acquiring voltage, current and power data of an electrical system; the electric fire early warning server is used for processing early warning information and calculating risk level of the acquired data; the central control system is used for controlling the operation of the whole early warning system; and the risk information display and push module is used for presenting, pushing and releasing the fire early warning information.

Preferably, the external environment parameter detection module and the electrical equipment parameter module both perform data transmission in a WIFI or RS485 network transmission mode.

Preferably, the risk information display and push module presents, pushes and releases the fire early warning information in an APP, short message or PC end audible and visual alarm mode.

Preferably, the external environment parameter detection module comprises a temperature sensor and a linear temperature sensor; the electrical equipment parameter module comprises a current-voltage transformer.

Preferably, the distribution positions of the temperature sensor, the linear temperature sensor and the current-voltage transformer are divided into areas according to the functional units of the tobacco distribution center.

An intelligent electric fire early warning method for a tobacco warehouse specifically comprises the following steps:

s1, collecting current, voltage, environment temperature and line temperature data of the electric fire early warning system;

s2, carrying out fuzzy comprehensive evaluation on the collected data to obtain a learning sample;

s3, constructing a BP neural network model, and dividing the learning sample into a training set and a testing set according to a proportion;

s4, inputting the training set into the BP neural network model for training, and performing effect test by adopting the test set to obtain a trained fire early warning model;

and S5, inputting the data collected in the S1 into a fire early warning model to predict the electrical risk level in real time, and generating an electrical risk level report.

Preferably, the S2 is specifically:

s2.1, determining an evaluation interval corresponding to a risk grade comment set and a risk grade of early warning according to a fuzzy comprehensive evaluation method;

s2.2, selecting the data acquired in the S1, generating a 5-cluster data set according to the condition that the mass center number is equal to 5 through a k-means clustering algorithm, and selecting a point set closest to the cluster center through a k nearest neighbor algorithm;

and S2.3, comprehensively considering factors of current, voltage, temperature and linear temperature based on multi-factor coupling, and acquiring high-risk and extremely high-risk data situations, namely learning samples, from the point set through a cable temperature rise loading coupling experiment.

Preferably, the risk level panel comprises an extremely low risk, a medium risk, a high risk, an extremely high risk; the evaluation interval corresponding to the risk level comment set is set as follows in sequence: 0-20, 20-40, 40-60, 60-80 and 80-100.

Preferably, the S4 is specifically:

s3.1, carrying out normalization processing on the training set data, and then inputting the processed training set into the BP neural network model for training;

s3.2, selecting a Sigmoid function as a BP neural network model activation function, and inputting the processed training set into the BP neural network model to perform initialization weight, forward transmission of hidden layer calculation and output layer calculation and reverse transmission of weight update;

and S3.3, performing effect test by adopting the test set, and finishing training when the output result of the BP neural network model reaches the acceptable level of the test set to obtain the trained fire early warning model.

Compared with the prior art, the invention has the beneficial effects that:

(1) monitoring and analyzing key parameters comprehensively and in real time: due to the adoption of multi-factor coupling comprehensive early warning, the inaccuracy of the traditional single-factor threshold is eliminated.

(2) Intelligently predicting the risk of fire: the system can comprehensively analyze the established electric fire comprehensive risk prediction model based on the fuzzy comprehensive judgment theory and the BP neural network technology according to the fire risk key influence factor data monitored in real time, determine the risk occurrence possibility and the risk consequence severity quantitative value, and finally determine the electric fire comprehensive risk level value. The system has self-learning and self-improvement functions and can achieve the purpose of risk prediction with high precision and high reliability.

(3) Multidimensional and multichannel fire risk early warning: the early warning information can be issued and pushed to personnel of different levels and ranges, such as enterprise leadership, security management layer, production management layer, field operation layer and the like. Advanced equipment such as smart phone APP and PC client sides are utilized to carry out network information transmission of fire early warning information, and meanwhile, a visual implementation tool is utilized to complete conversion from data to visual images.

4) The automatic emergency treatment disposal system can be organically integrated with the existing emergency treatment disposal system (such as automatic spraying, supersonic fire extinguishing and the like) of a tobacco enterprise, and timely and effective emergency treatment disposal can be carried out when an unforeseen accident occurs. The starting and the operation of the emergency system are controlled by the high-precision and high-accuracy warning information output by the system, and the purposes of quick response, scientific decision, ordered scheduling, efficient disposal and the like of various electrical safety accidents are realized.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.

FIG. 1 is a functional diagram of the system modules of the present invention;

FIG. 2 is a network connection diagram of an electrical fire warning server according to the present invention;

FIG. 3 is a diagram of a BP neural network structure according to the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

Example 1

Referring to fig. 1, the present invention provides an intelligent electric fire early warning system for a tobacco warehouse, comprising: the system comprises an external environment parameter detection module, an electrical equipment parameter module, an electrical fire early warning server, a central control system and a risk information display and push module;

the electrical fire early warning server is respectively connected with the external environment parameter detection module, the electrical equipment parameter module and the central control system; the risk information display and push module is connected with the central control system.

Generally, a fire detector refers to an apparatus capable of recognizing whether a fire is occurring. Common in the industry are: the residual current type and temperature measuring type electric fire monitoring detectors are mostly adopted. The residual current type electric fire monitoring detector can only detect leakage current, and can not effectively prevent the problems of short circuit, overload load and the like; in addition, the insulation of the circuit is deteriorated after the circuit is aged, so that leakage current is generated, and at the moment, unnecessary misoperation is often generated by the residual current type detector, and the accuracy of the system is influenced. The alarm principle of the temperature-measuring type electric fire monitoring detector is that an alarm is triggered when the collected temperature exceeds a set threshold value. Meanwhile, the monitoring system cannot reflect the comprehensive operation state of the electrical system, such as the correlation change and the coherent action of key factors such as voltage, frequency, current, temperature and the like. However, the principle of the method is basically that a threshold value method is adopted, when detected parameters (leakage current, electric wire temperature and the like) in a protected line exceed an alarm reference value, an alarm control signal can be sent out to indicate an alarm part, and the capability of early warning and fire risk prediction is unavailable.

After the generality of the electrical fire and the characteristics of the electrical fire in a tobacco warehouse place are comprehensively considered, the most representative current, voltage, environmental temperature and line temperature are selected as an index system for risk early warning, data such as the voltage, the current and the power of an electrical system are acquired through an electrical equipment parameter module, and data such as the environmental temperature and the line temperature of the electrical system are acquired through an external environmental parameter detection module.

In the early warning system, a WIFI transmission mode and an RS485 network transmission mode are adopted, collected data are transmitted to a fire early warning server, and early warning information processing and risk level calculation are completed, and fig. 2 is referred. The central control system is used for controlling the operation of the whole early warning system; and the risk information display and push module is used for presenting, pushing and releasing the fire early warning information.

Conventional fire alarm devices (e.g. Cerberus ECO FS18 from siemens germany) deploy only temperature and smoke detectors and the arrangement where they need to be installed is not entirely rational. In order to solve the problem, the invention determines a reasonable point distribution and monitoring scheme aiming at the actual situation of a place, the point distribution scheme and the framework of the distribution positions of the sensors with different currents, voltages, powers and environmental temperatures are designed according to the situation of a factory, and the distribution center is divided into areas according to the functional units of the tobacco distribution center: the key equipment which can cause production stop, large spread range and huge loss and damage due to the occurrence of an electrical fire is used as a monitoring unit, an area with high electrical fire risk and high capital density (such as an elevated reservoir area) is used as the monitoring unit, and the area with the same production process and process device characteristics is divided into the monitoring unit.

The invention also provides an intelligent electric fire early warning method for the tobacco warehouse, which specifically comprises the following steps:

s1, acquiring current, voltage, ambient temperature and line temperature data through the external environment parameter detection module and the electrical equipment parameter module;

s2, determining learning sample based on fuzzy comprehensive evaluation

S2.1, determining a comment set: according to a fuzzy comprehensive evaluation method, determining a risk grade evaluation set V of early warning, wherein the risk grade evaluation set V is { extremely low risk, medium risk, high risk and extremely high risk }, (because the risk is absolute and the safety is relative, the risk is inevitable as long as the electrical equipment is in an operating state, the extremely low risk and the low risk do not need to be alarmed, the medium risk informs an equipment administrator to pay attention continuously, the high risk alarms the equipment administrator, and the extremely high risk informs the equipment administrator and a plant area safety administrator), and in order to quantitatively reflect the result of the fuzzy comprehensive evaluation, an evaluation interval of evaluation corresponding to each grade is set to be {0-20, 20-40, 40-60, 60-80 and 80-100 };

s2.2, generating a comment set based on fuzzy comprehensive evaluation: selecting current, voltage, electric power, ambient temperature and line temperature of electrical equipment collected by a plant area, firstly generating 5 clusters of data sets according to the center of mass N _ clusters equal to 5 by a k-means clustering algorithm, then selecting typical data (namely a point set closest to a cluster center) in each cluster of data sets by a k-means clustering algorithm, inviting N experts in the field to score the risk level of each data in order to enable the score of a comment set to more objectively reflect the risk level of an actual electrical fire, and sequentially calculating the average value of the scores:

wherein, A is the average value of the scores of each group of data experts; n is the total number of experts;the score that the ith expert scored the set of data.

S2.3, as most of data collected on site are at medium risk and below, high-risk output cannot be obtained by training with the data, and in order to obtain the high-risk data, factors such as current, voltage, temperature, linear temperature and the like are comprehensively considered through laboratory experiments and based on multi-factor coupling, and high-risk and extremely high-risk data situations are obtained through cable temperature rise loading coupling experiments;

the resulting training data are shown in table 1:

TABLE 1

S3, constructing a BP neural network model, and dividing the learning sample into a training set and a testing set according to a proportion;

the structure of the BP neural network is shown in fig. 3.

The BP neural network mainly comprises an input layer, a hidden layer and an output layer.

And (3) normalization processing of data: the data input by the input layer is the data obtained in S1, and 80% of the data is used for training the BP neural network model, and 20% of the data is used for detecting the accuracy degree of the prediction of the test model.

S4, inputting the training set into the BP neural network model for training, and performing effect test by adopting a test set to obtain a well-trained fire early warning model;

in the BP algorithm, a Sigmoid function is generally selected as an activation function, and in order to accelerate the training and learning speed of a model, normalization processing needs to be performed on data:

wherein the content of the first and second substances,is the normalized data of the data obtained by the method,X_maxand X_minAre respectively a sample X_nMaximum and minimum values of.

Model training: inputting training data into a BP neural network model, respectively performing a forward transmission process of initialization weight-hidden layer calculation-output layer calculation and a reverse transmission process of weight updating, and completing model training when the testing effect of a 20% testing set reaches an acceptable level.

And S5, inputting the data collected in the S1 into a fire early warning model to predict the electrical risk level in real time, and generating an electrical risk level report.

The method comprises the steps that a model algorithm is deployed to a cloud server or a factory server room, monitoring equipment monitors current, voltage, electric power, ambient temperature and line temperature and transmits the current, the voltage, the electric power, the ambient temperature and the line temperature to an early warning model through a network, and then the risk level of a user can be informed in real time, so that the purposes of risk early warning and management and control are achieved, in order to obtain the electrical risk level of a region, the risk level of each unit (single electrical equipment) needs to be weighted and accumulated

Wherein R represents the overall risk level of the region, w_iWeight, R, representing electrical risk of each unit_iIndicating the risk level of each unit.

Besides, multi-dimensional and multi-channel fire risk early warning: the system can display the state parameter values and the risk level values of all the distribution positions and the comprehensive risk level judgment results of the electrical fire in real time, can derive the state parameter values and the risk level values as risk level reports, and can provide recommended measures for subsequent maintenance of companies and comprehensive risk level judgment results of the electrical fire. The multi-dimension mainly refers to issuing and pushing early warning information in a hierarchical and range-based manner, and if the levels of the predicted comprehensive risk levels are different, the early warning information can be issued and pushed to personnel of an enterprise leader, a safety management layer, a production management layer, a field operation layer and the like in different levels and ranges. The multi-channel refers to the network information transmission of fire early warning information by using advanced devices such as smart phone APP and PC clients; meanwhile, a visual implementation tool is used for completing the conversion from data to visual images, and on the basis of traditional characters and audible and visual alarm, visual and visual graphic demonstration, multimedia animation and other expression means are adopted, particularly, a geographic information system is used for carrying out accurate positioning means on a high fire risk area, and pushing and issuing of early warning information are carried out.

Emergency linkage: compared with the conventional fire alarm device which can only alarm and cannot be linked to eliminate the generated fire, the system can be organically integrated with the existing emergency treatment disposal system (such as automatic spraying, supersonic fire extinguishing and the like) of the tobacco enterprise, and timely and effective emergency treatment disposal is carried out when an unpredictable accident occurs.

In addition, the intelligent electric fire early warning system for the tobacco warehouse has an ideal field application effect, and can be deployed to more tobacco logistics industry companies, so that the fire hazard of the tobacco warehouse is practically reduced. The invention can be applied to the electric fire accident early warning of production plants and storehouses, can be widely applied to the electric fire early warning of office places, high-rise residential buildings and the like, and has wide application prospect and engineering value.

In conclusion, the invention achieves the following technical effects:

(1) monitoring and analyzing key parameters comprehensively and in real time: due to the adoption of multi-factor coupling comprehensive early warning, the inaccuracy of the traditional single-factor threshold is eliminated.

(2) Intelligently predicting the risk of fire: the system can comprehensively analyze the comprehensive risk prediction model of the electrical fire according to the key influence factor data of the fire risk monitored in real time and based on the established fuzzy comprehensive judgment theory and the BP neural network technology, determine the risk occurrence possibility and the risk consequence severity quantitative value, and finally determine the comprehensive risk level value of the electrical fire. The system has self-learning and self-improvement functions and can achieve the purpose of risk prediction with high precision and high reliability.

(3) Multidimensional and multichannel fire risk early warning: the early warning information can be issued and pushed to personnel of different levels and ranges, such as enterprise leadership, security management layer, production management layer, field operation layer and the like. Advanced equipment such as smart phone APP and PC client sides are utilized to carry out network information transmission of fire early warning information, and meanwhile, a visual implementation tool is utilized to complete conversion from data to visual images.

4) The automatic emergency treatment disposal system can be organically integrated with the existing emergency treatment disposal system (such as automatic spraying, supersonic fire extinguishing and the like) of a tobacco enterprise, and timely and effective emergency treatment disposal can be carried out when an unforeseen accident occurs. The starting and the operation of the emergency system are controlled by the high-precision and high-accuracy warning information output by the system, and the purposes of quick response, scientific decision, ordered scheduling, efficient disposal and the like of various electrical safety accidents are realized.

The embodiments described above are only for describing the preferred mode of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.